Error when using OLS trendline on single-point data

[slishak]

This could be an issue with statsmodels rather than Plotly, but when trying to fit trendlines to a line which only contains one repeated point, only one parameter is returned from the statsmdels fit and an error occurs when trying to build the hover header for the trendline.

import pandas as pd
import numpy as np
import plotly.express as px

df = pd.DataFrame({'y': np.zeros(2), 'x': np.ones(2)})
fig = px.scatter(df, x='x', y='y', trendline='ols')

Traceback (most recent call last):
  File "C:\Users\slishak\AppData\Local\Programs\Python\Python38\lib\site-packages\IPython\core\interactiveshell.py", line 3331, in run_code
    exec(code_obj, self.user_global_ns, self.user_ns)
  File "<ipython-input-17-ef61e6ef2734>", line 1, in <module>
    fig = px.scatter(df, x='y', y='x', trendline='ols')
  File "C:\Users\slishak\AppData\Local\Programs\Python\Python38\lib\site-packages\plotly\express\_chart_types.py", line 62, in scatter
    return make_figure(args=locals(), constructor=go.Scatter)
  File "C:\Users\slishak\AppData\Local\Programs\Python\Python38\lib\site-packages\plotly\express\_core.py", line 1907, in make_figure
    patch, fit_results = make_trace_kwargs(
  File "C:\Users\slishak\AppData\Local\Programs\Python\Python38\lib\site-packages\plotly\express\_core.py", line 298, in make_trace_kwargs
    fit_results.params[1],
IndexError: index 1 is out of bounds for axis 0 with size 1

Versions:

plotly==4.8.0
statsmodels==0.11.1

[nicolaskruchten]

Thanks for reporting this! I think this is a regression and we'll get it fixed shortly :)

[nicolaskruchten]

Ah, I spoke too soon: not a regression. But we'll still fix it :)

[nicolaskruchten]

Clarification: the problem here is that when the slope is zero, statsmodels returns only the intercept.